Podcast Summary: Patrick Collison on Stripe’s Early Choices, Smalltalk, and What Comes After Coding
Podcast: The a16z Show
Date: February 20, 2026
Host: Michael Truel (Cursor CEO, guest host for this episode)
Guest: Patrick Collison (CEO, Stripe)
Episode Overview
In this wide-ranging, reflective conversation, Stripe CEO Patrick Collison discusses the formative technology choices that built Stripe, the appeal and shortcomings of programming paradigms from Lisp and Smalltalk, the persistent challenges with developer productivity, and the prospects for programmable biology. Collison also explores the limitations and potentials of AI for software development, why foundational abstractions matter for business longevity, and offers thoughts on economic productivity and the diffusion of new technologies.
Key Discussion Points & Insights
1. Early Programming Experiences and Paradigms
[01:43]
- Smalltalk & Lisp Roots: Collison recounts building his first startup in Smalltalk after frustrations with mainstream frameworks. He praises Smalltalk’s interactive development, ability to debug and resume mid-execution, and its proximity to Lisp-like environments.
- “Smalltalk is actually this extremely interesting development environment that had a lot of the aspects of Lisp that I’d really appreciated there. Like a fully interactive environment with a proper debugger so that you can edit the code while in the middle of some request…” (Patrick Collison, 01:49)
- Hiring Non-Mainstream: Using non-mainstream languages like Smalltalk was not a hiring impediment because “smart people learn languages really quickly.” (Patrick Collison, 04:07)
- On Early AI Bots: Collison’s early AI experiment in Lisp for MSN Messenger relied on simple statistical modeling, not neural nets, and gave him formative experience with Lisp and genetic algorithms.
2. Enduring Lessons from Past Programming Ecosystems
[08:15]
-
Borrowed Ideas in Modern Languages: Features like interactive debugging from Lisp and Smalltalk have trickled into JavaScript environments, but Collison laments the loss of integrated development environments, where runtime, editing, and debugging are not separated.
- “I think the idea of…development environment and not just text editor is really the right idea. …That’s the thing that Lisp machines had…and Smalltalk has.” (Patrick Collison, 08:15)
-
Critique of Modern Paradigms: Collison calls for richer, insight-packed development environments—profiling, logging, error info—all fully integrated and overlaid onto code.
3. Theories on the Future of Programming
[12:51]
- Rise of Higher-Level Tooling: Truel and Collison agree that future coding will be less about syntax and more about intent, with AI facilitating higher-level abstraction and even changing what a “programming language” is.
- “Programming languages actually change and they can start to get a little bit less formal, …more about what you want and a little bit less about how you do it.” (Michael Truel, 13:06)
- Slow Experimental Change: Despite vast potential, Collison notes that programming paradigms have seen little revolutionary experimentation since the 1980s, especially in development environments.
4. Stripe's Early Technology Choices and Their Lasting Impact
[21:44]
- APIs, Data Models, and Conway’s Law: Collison stresses that APIs and data models fundamentally shape not just codebases, but company strategy and organizational structure.
- “I think the strong version [of Conway's Law] is that it substantially shapes your strategy and just your business outcomes.” (Patrick Collison, 18:46)
- Why Stripe Picked MongoDB & Ruby: A principled objection to SQL’s conceptual mismatch led to choosing MongoDB; Ruby was selected for being mainstream enough, but still flexible.
- “I just had this principled objection to SQL…so instead of using Smalltalk, okay, we weren’t going to go to Java, but we went to Ruby… and used Mongo, which still give a lot of flexibility…” (Patrick Collison, 25:38)
- Living with Early Decisions: Some technologies chosen at Stripe’s founding—MongoDB, Ruby—still form core infrastructure. This reflects both the lasting impact of “Big Bang” startup decisions and the challenge of later migration/change.
- “Stripe is now 15 years old. And there are lots of things that we designed 15 years ago that are still in use today, which is kind of good and bad…” (Patrick Collison, 18:46)
5. Stripe API V2 Rebuild: Lessons from Deep Refactoring
[27:20]
- Rewriting Core Abstractions: Realizing that previous abstractions had long-term shortcomings, Stripe is rolling out V2 APIs, aiming for unification and better modeling of business realities.
- “We are unifying all of those into being…the same kind of entity representation, which is on some level clearly the right answer and…is already changing the businesses of some of our customers…” (Patrick Collison, 28:29)
- Complex Upgrades: Introducing V2 is likened to an “instruction set migration”; defining new APIs is easy, but making them interoperable and ensuring sensible migration paths are hard (Patrick Collison, 28:29).
- API Design Process: Strong leadership is essential—one person must deeply own the whole. Early customer feedback and hands-on use-cases prevent overengineering.
- “There is a working group…but there is also a singular person who understands and is more than anyone else responsible for the whole. And I think that's necessary.” (Patrick Collison, 30:59)
- “I think the cycles of customer validation, customer feedback are…extremely important.” (Patrick Collison, 32:18)
6. Reflections on Productivity, AI, and The Economy
[36:47]
- Has AI Moved Productivity Yet? Despite AI’s promise, Collison observes no clear GDP “takeoff,” pointing to recent studies that find little measurable lift in productivity from LLMs so far.
- “Its claim is that one does not in fact observe productivity improvements stemming from…language models…” (Patrick Collison, 40:15)
- “We’re not living in some, you know, massively accelerated period of economic growth for the world writ large …diffusion… takes time and involves substantial complexity.” (Patrick Collison, 41:10)
- Will We Need New Economic Metrics? Collison argues GDP and traditional measures are robust enough; if real effect emerges, it will eventually show “in the numbers.” (Patrick Collison, 43:06)
7. Programming Biology: The Next Turing Loop
[43:25]
- Programmatic Biology: At ARC, Collison is working to create foundational models for biology—reading (sequencing), thinking (AI models), and writing (CRISPR/editing). The ultimate goal is to make progress against complex diseases, a problem previously unsolved due to their combinatorial complexity.
- “If you put those together, you now have the ability…to read, think and to write. And this starts to really feel like a new kind of Turing loop and to have its own sort of completeness.” (Patrick Collison, 46:10)
8. AI and The Changing Skillset for Builders
[47:02]
- Who Benefits if Programming Is Automated? Collison resists speculation, noting the surprising, sensitive nature of technological shifts. Traditional economic doctrines may not predict winners; many “stock answers” (real estate, commodities, designers, grad students) could be right or wrong.
9. How Can AI Tools Like Cursor Improve?
[50:18]
- Three Suggestions for Code Tooling:
- Deeper Integration: More runtime characteristics, profiling, and error information woven into the experience.
- AI-powered Refactoring: Tools that improve code structure and maintainability automatically.
- Upholding Craft: Ensure AI enables not just more code, but more beautiful, well-crafted, reliable software.
- “We care about…craft and beauty. We want our software to be well designed…There’s obviously a concern with AI that it leads to the creation of more slope and more kind of crappy things, but not more of the best things.” (Patrick Collison, 50:18)
Notable Quotes
-
“The basic idea of as development environment and not just text editor is really the right idea. And that’s the thing that Lisp machines had…That’s the thing that to some extent Mathematica has. That’s the thing that Smalltalk has. And I think it’s just such a mistake that we have ended up with development environments where there is such a separation…” (Patrick Collison, 08:15)
-
“If I was to do everything at Stripe again… the thing that I think we could maybe foreseeably and beneficially have done differently would be to have spent even more time than we did on APIs and data models…” (Patrick Collison, 18:46)
-
“For none of [the complex diseases] can we really say that we’ve cured it, that we understand the causal pathways in meaningful detail…Our hypothesis…is that this is in part because we don’t have experimental…and epistemic technology that’s up to the task…” (Patrick Collison, 43:27)
Timestamps – Important Segments
- [01:43] Smalltalk and Lisp experiences, importance of interactive/deep programming environments
- [08:15] Reflections on integrated environments and what modern tools can learn from the past
- [18:46] Stripe’s API/data model choices: long-tail impact and organizational shape
- [27:20] Stripe’s V2 API rewrite: rationale, process, and lessons
- [36:47] Does AI show up in productivity stats? Economics, diffusion, and real-world impact
- [43:25] Programming biology: ARC’s mission for biomedical “Turing completeness”
- [50:18] AI tooling wish list: deep integration, AI-empowered refactoring, and building software “craft and beauty”
Memorable Moments
- Tech Choices Aren’t Just Tech: Collison explains how early technical decisions—frameworks, languages, database models—become foundational, shaping a company’s very culture and trajectory.
- APIs as Strategy: Drawing parallels between iOS/Android ecosystems and Stripe’s own history, Collison underscores that abstraction design carries business consequences decades out.
- Caution on AI Overoptimism: Noting slow macroeconomic uptick, Collison urges critical thinking, not magical thinking, about productivity leaps from AI.
Conclusion
This episode delivers a rare, grounded mix of technical nostalgia, practical takeaways, and future-looking vision from one of tech’s leading founder-CEOs. Patrick Collison’s candor about Stripe’s historical decisions, current API overhaul, and cautious optimism about AI and biology sets this episode apart for technologists, business leaders, and progress enthusiasts alike.
